Method and apparatus for process optimization

This application is continuation of U.S. patent application Ser. No. 16,589,713 filed on Oct. 1, 2019 which in turn is related to and claims the benefit of U.S. Prov. Applications entitled: (1) “Method and Apparatus for Process Optimization” which was filed on Oct. 1, 2018 which received U.S. Provisional Application Ser. No. 62/739,441; and (2) “Method and Apparatus for Process Optimization” which was filed on Feb. 4, 2019 and which received U.S. Provisional Application Ser. No. 62/800,900. All of these applications are incorporated in their entireties herein by reference for all purposes.

Any external reference mentioned herein, including for example websites, articles, reference books, textbooks, granted patents, and patent applications are incorporated in their entireties herein by reference for all purposes.

Executing a process defined by a protocol or procedure is part of virtually every modern industrialized field. For example, scientists execute experimental protocols, health care providers execute clinical protocols, factory workers execute manufacturing procedures. For a successful outcome, one must execute the steps of such protocols and processes in a reproducible and repeatable manner. However, there are challenges to doing so. Since these processes are executed in the real physical world, and often by humans, there are myriad variables that can introduce errors and variations to lessen reproducibility and repeatability. These reproducibility problems are well known in industry and as a result there have been various methods developed to overcome, prevent, or address such process variations, such as the “Six Sigma” method, among others.

However, such methods are costly to implement across a facility and can take a very long time to implement. Accordingly, implementation of quality control/standardization methodologies typically are more feasible and typically more often applied in large facilities and processes. There is a need for a more flexible solution that can be used in both small-and large-scale facilities and processes.

In a first embodiment, the present invention provides a method for identifying (e.g. calculating/determining/observing/saving/estimating/creating etc.) a preferred trajectory model(s) for achieving successful execution of a process which has a defined protocol (e.g. steps/structure) to perform during the process. The method includes the steps of:

In another embodiment, the present invention provides a method for identifying a preferred trajectory model(s) for achieving successful execution of a process, the process having a defined protocol to perform during the process, the method comprising the steps of

In another preferred embodiment, the present invention provides a method for determining whether a given run of a process having a defined protocol is on a trajectory for successful completion, the method comprising the steps of:

In other embodiments, the present invention provides a set of written instructions, a computer, a computer program, a software package, a module and/or a node programed with logic and/or instructions for performing any and/or all steps of any method of the present invention.

In additional embodiments, the present invention provides a successful process trajectory data file (and a computer containing such a file), wherein the file includes a plurality of value sets of variables that affect the outcome of a process compiled from various individual runs of a given process.

The present invention provides solutions to the above-described problems in the art.

In one embodiment, the present invention provides a method for determining the trajectory of success of a process over the course of performing a protocol or defined steps to complete the process. Upon completion of the process a result is achieved such as the production and/or analysis of a material and/or sample.

The method includes steps of obtaining or measuring variables related to the process over the course of performing the process. The variables include those that may affect the quality of the process and/or that may affect the result achieved upon completion of the process (e.g. quality of the resulting material produced and/or quality of analysis, etc.).

The variables can include “n” number of variables that may affect the quality of performing the process and/or the outcome/result achieved by the process and are not limited herein. For example, the variables can include, among others:

Some of these variables, for example those related to environmental conditions, can be measured/determined using associated environmental sensors (or sensor packages/units) placed on or near equipment used in the process (e.g. laboratory and/or manufacturing equipment). One such particularly preferred environmental sensor package is described in U.S. Prov. Application entitled “Method and Apparatus for Local Sensing” which was filed on Oct. 1, 2018 and received U.S. Provisional Application Ser. No. 62/739,419 (which is incorporated herein by reference). This application describes a label/tag sensor package comprising a plurality of sensors configured on a small flexible backing for local sensing applications. This smart label sensor package can be placed on laboratory/manufacturing equipment, storage containers, and even on products and/or packaging as the product is produced, stored and/or shipped. This sensor package can measure/determine many of the environmental factors of interest and described herein and can wirelessly communicate this data to an application server for aggregating with measurement data received from process instruments in the methods herein described. Furthermore, due to the size and relatively low cost of these sensor packages, they can be placed at many different locations (e.g. such as on tools and instruments) within a facility and measure local environmental conditions with ease, etc.

As the protocol/steps of the process are executed, the values of these variables can be measured as a function of time over the course of the process. The values of these variables are determined and preferably transferred to and/or otherwise stored in a in a file system (e.g. such as one having optical and/or electronic storage means in a file structure and/or file hierarchy, such as a database, etc.) (e.g.g. resident in the facility or remote server via the internet). The database can be for example known empirical data management systems (EDMSs such as a electronic lab notebook (ELN) and/or scientific data management system (SDMS) and/or laboratory information management system (LIMS) etc.). The values of these variables are preferably transferred/stored in the database along with physical values and/or other information and/or data obtained and received during execution of the steps/protocol of the process. For example, if a step of the protocol of the process requires adding a certain amount of material to a beaker or reactor, etc. at a certain time, the value associated with measurement of the material (e.g. the scale reading) along with other variables that may affect the reading or the step such environmental conditions such as the temperature, humidity, light conditions, etc. of the environment where and when the material measurement was made can also be determined and recorded along with the actual value of the physical measurement.

One particularly preferred way to accomplish these tasks is described in U.S. provisional application entitled “Systems and methods to integrate environmental information into measurement metadata in an Electronic Laboratory Notebook Environment” which received U.S. Provisional Application Ser. No. 62/739,427, and which is incorporated herein by reference for all purposes. In this reference, exemplary variables which may affect the process or outcome thereof (e.g. including environmental variable data), is obtained and transferred/stored/aggregated (e.g. preferably as metadata) with physical data relating to the given process step or other measurement.

As the process is performed repeatedly (e.g. the protocol and steps of the process are executed repeatedly), the values of the variables that may affect the quality and/or outcome of the process can be measured as a function of time and transferred or otherwise saved in an empirical data management system (EDMS) such as an electronic lab notebook (ELN). This provides an n-dimensional trajectory map of each variable's value over time for each process run as the process is executed (if plotted as a function of time, it results in an n+1 dimensional representation with n variables plus time).

The output of the quality of each process run (e.g. via quality control testing etc.) is quantified. For example, each process run can be quantified by the result (e.g. quality of the product and/or quality of the analysis). In particular, did the particular run create a product having preferred properties, acceptable properties, or unacceptable properties, etc.

The output of the quality of the process run can then be applied to each trajectory thereby providing a weighted trajectory map based upon the quality of outputs of the associated trajectories/runs.

As more runs are executed, each trajectory in n-dimensional space can continue to be weighted by the quality of the output (that is by the appropriate QC metrics) and mapped. Weighting can take into account the importance of each step in the process (i.e. each step can be weighted differently depending on its importance a priori). Weighting can be done automatically by correlating which process steps are more correlated with the quality of the outcome.

An “ideal/preferred trajectory” of any given process (or threshold values for an ideal or acceptable trajectory) can be determined based on appropriate statistical treatment of the trajectory population (can be mean, median, weighted average, etc) but should be weighted by the quality metrics.

Once an ideal/preferred trajectory is determined for the particular process at hand (and preferably stored as a file), future runs of that process can be compared against it to determine how “close” that particular future run is to the ideal process trajectory. “Closeness” of the particular run can be computed in many different ways, including the Euclidean distance, as well as other known techniques for determining “closeness” or similarities of curves in n-dimensional space.

Use of the measure of “closeness” to an “ideal/preferred trajectory” can indicate a probability of success of a given run of the process at any given step of the process. In particular, when a process is running, a user can tell in real time whether or not it is on the right path. If the run starts to deviate from the “ideal trajectory” then the user can be provided with information and/or instructions to either stop the process (which can save cost) or can alter the process appropriately to get it back on track, or let the process proceed and then mark the output product for special handling as needed (for example, it can be discarded or quarantined). Once the user knows the likelihood of success of the executed run of the process, the user can decide (or be provided with instructions) whether to continue, or not to continue, with the output product in subsequent process steps. This can save time/money/effort/materials/etc.

The measure of “closeness” can be applied and/or ascertained via use of a computer implemented program/algorithm stored on a user's computer, local network, or external network or server. The program can make use of information and data received to the variables from the current process run and compare the received variables against the stored “ideal/preferred trajectory” to provide a calculated measure of “closeness” of the current run to the “ideal/preferred trajectory”. The computer/program/module/node etc. can also be programmed with logic/instructions to compare the measured value of closeness with a lookup table or similar value or file, to provide an indication of the likelihood of successful completion of the current run. The computer/program/module/node may likewise be programmed with logic/instructions to perform a cost benefit analysis with respect to the determined likelihood of success of the current and whether it is economically feasible to continue the current run give the likelihood of success of the run, etc.

As can be seen throughout, the present invention provides new and useful methods for identifying (e.g. calculating/determining/observing/saving/estimating/creating etc.) a preferred trajectory model(s) for achieving successful execution of a process. The process can be any time of process (e.g. laboratory, manufacturing, human action process) which has a defined protocol (e.g. steps, structure, architecture) to perform during the process to complete at least a portion of the process.

The method includes a first step of initiating one of more runs of the process and preferably running until at least one result metric is determined (or more preferably to the process is complete). This can be running the process until one of the process steps of the process are completed or otherwise running until some result metric can be determined from said initiated run. In other embodiments, the process protocol is run to completion in the initiated run where all protocol/steps/structure of the process are performed during that run. In short, the initiated run is allowed to proceed such that a result metric can be obtained and used to compare that run to other initiated runs or parts of initiated runs to that metric (e.g. for example the result metric indicates whether a run is on track: for success or failure; to produce an acceptable/preferred/bad/OK product or results or analysis; etc.). In most preferred embodiments several runs are initiated and run to at least a point where result metrics from those runs can be obtained and used to compare to other of the initiated runs as described.

During the one or more runs, or parts of runs, initiated in the first step, information is obtained (e.g. measured/obtained/received) which reflects one or more variables that may affect the quality of said run and process. This occurs most preferably during an initiated run that is obtains a favorable result (e.g. considered successful run) and also preferably during runs that obtain unfovarable results (e.g. considered unsuccessful) and most preferably during all runs whether they are successful or not. The outcomes of each run produce variable outcomes and useful information, wherein the different values of the variables that may affect the quality of the process as a function of time are determined for each run.

In a next step a trajectory path of the variables obtained in the previous step are represented as a function of time for each run as a separate trajectory path. For example, this could be accomplished by creating a mathematical representation via modeling, plotting, three dimensional vectors, multi-dimensional arrays, and/or tensor.

The outcome of each run (e.g. whether the run achieved a successful or unsuccessful result) is then determined (e.g. the run is scored/evaluated, etc.) and then each trajectory path is weighted (e.g. scored etc.) according to the achieved outcome (e.g. perfect/ideal result, preferred result, goods acceptable result, unacceptable/fail), to create a weighted representation of the initiated runs (e.g. plot). From this weighted representation (e.g. plot) a preferred/ideal process trajectory model can then be identified (e.g. calculated, determined, observed, saved, estimated etc.) based upon analysis of the weighted representation (e.g. weighted plot). In preferred embodiments, a record (e.g. file/record/receipt/note etc.) regarding the identified ideal trajectory and/or the associated variables which led to the ideal trajectory is generated and/or stored, for use in later comparison of subsequent process runs against the identified ideal trajectory, etc. When the trajectory path of later process run is compared to the ideal/preferred trajectory, it can be determined whether the later process run is on a path for success or failure and depending on this determination, it can be determined whether the later process run should be continued (if on a likely path of success) or whether the later process run should be discontinued (if on a likely path for failure) to save resources, whether the process should be modified in order to put the run on a trajectory for success, and/or whether to institute a new run.

The variables that may affect the quality of an initiated process or run are not particularly limited. In some embodiments, the variable may include any one or combination of:

In another embodiment, the present invention provides a method for determining whether a given run of a process having a defined protocol (e.g steps/structure/recipe/architecture) is on a trajectory for successful completion. The method comprising the steps of:

In preferred embodiments, wherein upon performing comparison step (d) it is determined that the run initiated in step (a) is on a path or trajectory for failure, the process further comprises the step of abandoning or providing instructions to abandon the run initiated in step (a).

Conversely, in other preferred embodiments, wherein upon performing comparison step (d) it is determined that the run initiated in step (a) is on a path or trajectory for success, the process further comprises the step of continuing the process run or providing instructions to continue the process run initiated in step (a).

In additional preferred embodiments, wherein upon performing comparison step (d), it is determined that the run initiated in step (a) is not on a path or trajectory for success, the process further comprises the step of adjusting/modifying/altering or providing instructions for adjusting/modifying/altering the process protocol or steps to improve process performance and/or put the run initiated in step (a) on a path or trajectory for success.

In preferred embodiments herein described, the methods and system preferably make use of computer/program/module/node/infrastructure programmed with logic/instructions and having circuity comprised of hardware, software, memory, processors, data storage, computers, etc. which cause/create/effect operability of said systems and methods. In these embodiments, the present invention provides a successful process trajectory data file which comprises numerous value sets of variables that can affect the outcome of a process compiled from various individual runs (or parts of runs) of a given process and a computer, server, data storage facility comprising such a file or logic or instructions containing such a file. Furthermore, and in other embodiments, the present invention provides a printed set of instructions and/or a computer program/module/node programmed with logic and/or instructions performable by a computer processor to perform any method herein described.

In further preferred embodiments, any of the methods and/or steps of the present invention are preferably performed using the EDMSs (e.g. electronic lab notebook (ELN) systems and/or aggregated data systems) and/or methods as described in U.S. Prov. Application entitled “Systems and methods to integrate environmental information into measurement metadata in an Electronic Laboratory Notebook Environment” which received U.S. Provisional Application Ser. No. 62/739,427, and which is incorporated herein by reference for all purposes.

Reference throughout the specification to “one embodiment,” “another embodiment,” “an embodiment,” “some embodiments,” and so forth, means that a particular element (e.g., feature, structure, property, and/or characteristic) described in connection with the embodiment is included in at least one embodiment described herein, and may or may not be present in other embodiments. In addition, it is to be understood that the described element(s) may be combined in any suitable manner in the various embodiments.

Numerical values in the specification and claims of this application reflect average values for a composition. Furthermore, unless indicated to the contrary, the numerical values should be understood to include numerical values which are the same when reduced to the same number of significant figures and numerical values which differ from the stated value by less than the experimental error of conventional measurement technique of the type described in the present application to determine the value.

shows two runs of a 4-step process. As can be seen, each of the four steps are performed sequentially. In each of the two runs, the steps take different amounts of time to perform, and the time between steps is different. These variations may be due to personnel differences, availability of equipment, among many other reasons.

In each step, there may be different variables that can affect the quality of the run. For example, these variables may include: the temperature and humidity in the room when each step was executed; the level of particulates in the room when each step was executed; the time elapsed since an instrument used in a step was last calibrated; etc. Accordingly, each step of the process has multiple dimensions of variables that can affect the quality of that step's output.

A materials science company is formulating a complex material. There are many steps in the process, with each step adding more value (and thus more cost) to the material as it is formulated. Unfortunately, the material cannot be tested in a non-destructive manner, so the best option available is to sample the batch and destructively test a representative sample. There is a need for a way to determine the quality of the material as it is being formulated that is better than destructive testing of a sample of the batch.

In this example, step 1 involves incubating a polymer material in an oven and therefore the temperature profile over the course of step 1 is a relevant metric for quality.shows temperature versus time for multiple runs of step 1. Each curve represents the temperature that the polymer material experienced as a function of time over the course of given process run.

The set of temperature curves indicated by the dotted circleall resulted in high quality polymeric material, as determined by QC testing results. Furthermore, the other temperature profiles (-) resulted in poor quality material. Thus it is desirable for future runs to be close to the runs indicated by circle.

shows an example of how the desirable curves in() can be represented as a single curve. Curveis the ideal temperature profile curve based on the multiple curvesin. Curvemay be determined by different methods, but in one example, it may be determined by averaging the set of curvessince all of the curvesresulted in high quality polymer material as determined by their QC testing.

Furthermore, since curves-inresulted in lower quality polymer, the distribution of these curves can be used to compute boundaries that describe the expected quality of the polymer. For example, curves that fall betweenandmay be expected to pass QC testing 99% of the time; curves that fall betweenandmay be expected to pass QC testing 90% of the time; curves that fall betweenandmay be expected to pass QC testing 80% of the time.

reproduces the same curves as shown in. The ideal run curve is, the 99% boundary curves areand; the 90% boundary curves areand; the 80% boundary curves areand. The company then performed a subsequent run, as shown by curve. As can be seen, this run falls between curvesand, indicating a 80% likelihood that the polymer produced by this run will pass QC testing. In this case, the company may elect to discard this material at the end of Step 1 and not continue to incur costs associated with implementing Steps 2-4 on this particular run of polymer, thereby saving the company money, time, and effort if in fact the run is not successful in producing a polymer that meets QC.

illustrates an example where the trajectory of a rundoes not fall cleanly within one of the predetermined bands. In this case, one can estimate the quality of the output by determining the closeness of the runto the ideal curve. Closeness can be determined by many ways that are known in the art, including Euclidean distance, cross correlation, area between curves, and also the method described by Vlachos, et al: http://alumni.cs.ucr.edu/˜mvlachos/pubs/icde02.pdf

This method can be generalized to any of the multiple variables that may affect the process as described above (or any combination thereof), not just temperature versus time. As such, a multi-dimensional trajectory map of how a process was executed can be created. Each curve in that multi-dimensional space can then be scored (or weighted) by the results of QC testing, thereby delineating an ideal set of conditions over time that need to be met to produce high quality output on a consistent basis.

Thus this invention describes a method for estimating the outcome of a process based upon the quality of adherence to a protocol and/or by deviation from a protocol, etc. Knowing the likelihood that a process (or part of a process) will yield a good result allows a user to take action to reduce costs, time, and effort. Furthermore, for processes where non-destructive testing of an intermediate product is not possible, such estimation techniques can give better insight than a sample-based approach since the latter only tests a small percent of the batch.

In brewing, the first step of the process is called ‘mashing’. During mashing, dry, milled malt is mixed with hot water to extract starch molecules and enzymes from the grain. The starch molecules are processed primarily by two enzymes, alpha-and beta-amylase, converting the starch molecules into a variety of fermentable and non-fermentable sugars. Alpha-amylase is an endo-enzyme that quickly reduces the starch size creating complex sugars, while beta-amylase is an exo-enzyme that primarily creates maltose molecules. Alpha-and beta-amylase operate optimally at two separate temperatures; alpha: 158F and beta: 131-149F. Controlling these temperatures during the mashing process ultimately determines extract yield (amount of sugar extracted from the grain) and fermentability (% of sugars that can be fermented by brewing yeast).